The present invention generally relates to on-line processing systems and overload suppressing methods, and more particularly to an on-line processing system and an overload suppressing method for automatically suppressing an overload of a buffer which is used as a storage region for temporarily storing messages.
In computer systems, there are continuous demands to increase the processing speed. Hence, when transferring a message via a buffer in an on-line processing system, there are demands to stop the work with a low priority before a buffer overload occurs and causes the work with a high priority to slow down, so as to carry out the work process with the high priority without hindrance. It is possible to detect the lack of buffer region (or space) to a certain extent by use of an application program which is made by the user. However, there is no means of detecting the work with the low priority, and it would be very useful if it were possible for the system to automatically detect the work with the low priority, that is, the work which does not need to be carried out with a high priority.
FIG. 1 is a diagram for explaining an example of a conventional on-line processing system. In FIG. 1, work processing parts 11a and 11b carry out predetermined work processes by application programs. The work processing parts 11a and 11b are coupled to an on-line control part 71. This on-line control part 71 includes a message control part 72 for controlling transfer of messages to and from terminal equipment and the work processing parts 11a and 11b, and a buffer control part 73 for managing a buffer (not shown) which is used as a storage region for temporarily storing the messages. In other words, the buffer control part 73 acquires and releases the buffer. A center console 21 is coupled to tile work processing part 11b. A buffer supervising application program 70 is made by the user.
The buffer control part 73 of the on-line control part 71 has a plurality of buffer pools, and manages the correspondence between a buffer which is selected from the buffer pools and each work processing parts 11a and 11b. If a large number of buffers are simultaneously used and it becomes impossible to secure a sufficient buffer region for storing input/output messages or the like, an overload occurs and the service of the work processing parts 11a and 11b stops. In this case, it is desirable to continue the process of the work processing parts 11a and 11b having a high priority and to stop the process of the work processing parts 11a and 11b having a low priority.
For this reason, the user conventionally makes the buffer supervising application program 70 for detecting the lack of buffer region, and detects the insufficient buffer region by executing this buffer supervising application program 70. The overload is suppressed by urging the operator to input an instruction to stop the work which is using the extremely limited buffer region.
In other words, the following process is carried out by the buffer supervising application program 70.
A step S1 shown in FIG. 1 sets a predetermined buffer supervision time for periodically supervising the buffer, and makes a supervision request to an operating system for this predetermined buffer supervision time. A step S2 waits for a time lapse notification from the operating system. A step S3 calls a buffer region (or amount) notification service provided by the buffer control part 73 if the predetermined buffer supervision time elapses, so as to investigate the present amount of buffer region used. A step S4 decides whether or not the free or available buffer region is insufficient. If the free buffer region is sufficient and the decision result in the step S4 is NO, the process returns to the step S1 so as to similarly continue the time supervision.
On the other hand, if the free buffer region is insufficient and the decision result in the step S4 is YES, a step S5 outputs to the center console 21 a message which indicates that the free buffer region is extremely limited and that the possibility of buffer overload is high. The process ends after the step S5. Accordingly, if such a message indicating that the possibility of buffer overload is high is output from the center console 21, the operator stops the work processing part 11b, for example, which is regarded as having the low priority, so as to eliminate the possibility of buffer overload.
According to the conventional on-line processing system shown in FIG. 1, the buffer region is supervised for every predetermined time, and it is thus impossible to detect the lack of buffer region in real time. As a result, even if the operator stops the work with the low priority when the lack of buffer region is detected, it may be too late and the buffer allocation with respect to the work with the high priority may fail.
In other words, even if the buffer supervising application program 70 detects the lack of buffer region and the work with the low priority is stopped, the buffer region may not be released in time, and there was a problem in that the process of the work with the high priority may slow down due to an insufficient buffer region.